www.nature.com/scientificreports

OPEN Metabolomics‑based analysis of miniature fask contents identifes tobacco mixture use among the ancient Maya Mario Zimmermann1*, Korey J. Brownstein4,5, Luis Pantoja Díaz2, Iliana Ancona Aragón2, Scott Hutson3, Barry Kidder3, Shannon Tushingham1 & David R. Gang4

A particular type of miniature ceramic vessel locally known as “veneneras” is occasionally found during archaeological excavations in the Maya Area. To date, only one study of a collection of such containers successfully identifed organic residues through coupled chromatography–mass spectrometry methods. That study identifed traces of nicotine likely associated with tobacco. Here we present a more complete picture by analyzing a suite of possible complementary ingredients in tobacco mixtures across a collection of 14 miniature vessels. The collection includes four diferent vessel forms and allows for the comparison of specimens which had previously formed part of museum exhibitions with recently excavated, untreated containers. Archaeological samples were compared with fresh as well as cured reference materials from two diferent species of tobacco (Nicotiana tabacum and N. rustica). In addition, we sampled six more plants which are linked to mind-altering practices through Mesoamerican ethnohistoric or ethnographic records. Analyses were conducted using UPLC-MS metabolomics-based analytical techniques, which signifcantly expand the possible detection of chemical compounds compared to previous biomarker-focused studies. Results include the detection of more than 9000 residual chemical features. We trace, for the frst time, the presence of Mexican marigold (Tagetes lucida) in presumptive polydrug mixtures.

Te induction of altered states of consciousness (ASC) is a common feature of ­humankind1, among hunting and gathering ­communities2 as well as complex ­societies3,4. ASCs cause either a heightened awareness of the senses or the perception of alternate realities. How these states are achieved varies profusely between cultures, yet there are two principal pathways: the creation or activation of pathomorphic stressors (through, e.g., fast- ing, sleep deprivation, bloodletting), which lead to the liberation of endogenous opioid peptides (endorphins)5, and the administration of entheogenic substances of either plant or animal ­origin6. Either way, many societies have specialists, such as shamans, traditional healers, and doctors, who help community members in achieving ASC­ 7. Moreover, McClenon­ 8 argues the susceptibility to shamanic/hypnotic suggestion was positively selected in humans and their ancestors for the evolutionary benefts of pain reduction, enhanced healing, and easier childbirth. Among entheogenic techniques, smoking in pipes is the most likely to be recognized archaeologically. How- ever, although Pleistocene hunter-gatherers settling the Americas from Siberia arguably brought along a complex of shamanistic practices and medicinal plant use­ 9, “nicotine delivery devices” (NDD) are an innovation not appearing in the North American archaeological record before the Late Archaic. Te earliest known dates for the US southwest and northeast fall within the frst millennium BCE­ 10, while to this day the oldest fnds from the Columbia River Basin reach back only 1200 years11. Tobacco (Nicotiana sp.) is the most prolifc producer of ­nicotine12, but this metabolite has been detected in lower concentrations in other ­genera13,14. Nicotiana species also contain the psychoactive harmala alkaloids harman and norharman, which are known for their efects in

1Department of Anthropology, Washington State University, P.O. Box 644910, Pullman, WA 99164‑4910, USA. 2Centro Regional Yucatán, National Institute of Anthropology and History, Col. Gonzalo Guerrero, Calle 10 #310‑A, 97310 Mérida, Yucatán, Mexico. 3Department of Anthropology, University of Kentucky, 211 Laferty Hall, Lexington, KY 40506‑0027, USA. 4Institute of Biological Chemistry, Washington State University, P.O. Box 646340, Pullman, WA 99164‑6340, USA. 5Present address: Department of Molecular Genetics and Cell Biology, The University of Chicago, Cummings Life Science Center, 920 E. 58th Street, Suite 1106, Chicago, IL 60637, USA. *email: [email protected]

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 1 Vol.:(0123456789) www.nature.com/scientificreports/

drugs like ayahuasca15. Even though tobacco was widely documented as both a medicine and a sacred plant among Native Americans­ 16,17, it was merely part of a much larger complex of psychoactive products, including plants that were smoked and that single-handedly account for dozens of genera­ 18. Terefore, archaeological research of the co-evolution of humans and drugs requires thorough methodological foundations and state-of- the-art technological tools. In this paper, we present a metabolomics-based case study from Mesoamerica with a focus on tobacco preparations as a window into the modes of drug consumption. In Mesoamerica, native communities such as the Nahua and Maya feature ritual specialists or shamans who communicate with supernatural entities and ancestors through ASCs induced by ingesting psychoactive plants as well as other pathways. De la ­Garza19 states tobacco is likely the most important sacred plant for the ritual and daily lives of Mesoamerican people. Interestingly, she specifcally refers to N. rustica rather than the more commonly known N. tabacum, which forms the base of present-day commercial products. With access being socially restricted, ASC-inducing plant products were most likely kept in specialized containers in the past. De la ­Garza19 argues the venenera fasks of the Maya Area are a strong candidate for such artifacts. Although the vernacular term venenera translates to “venom bottle”, these vessels have also been said to contain poison, medicine, pigment, or snuf. Troughout this article, we follow Loughmiller-Cardinal and ­Zagorevski20 in not making a priori assumptions about vessel contents and apply the neutral term “miniature fasks”. While information on storage containers is scarce, ethnohistoric sources from Mesoamerica provide more background on delivery devices and the mind-altering products themselves. Friar Bernardino de Sahagun was the frst outside observer to give detail on indigenous drug consumption in what is today central Mexico. He describes smoking products among contact-period Nahua in the following words: Tere are many ways of these canes, and they are made from many and diverse fragrant herbs, ground and mixed with each other, and flled and packed of roses, of fragrant spices, of the bitumen known as chapuputli, and of mushrooms, and of roses called poyomalli, or of tlzyetl, which is an herb. Sahagún 1999 [1569]:574. Based on ethnographic observation­ 21, Mexican marigold (Tagetes lucida) fgures notably among these tobacco mixture ingredients. In Mexico and Guatemala, this annual herb is commonly known for its role in ceremonies for the dead, which appear to have pre-Columbian ­roots22. However, present-day Huichol communities in west- ern Mexico continue to smoke dried leaves and fowers either by themselves or in a mixture with N. rustica21. Similar to most of their Mesoamerican neighbors, the ancient Maya did not use stone or clay pipes but other, more degradable NDDs such as reeds and corn husks­ 23. However, smoking was not the only way tobacco was consumed by the indigenous groups of Mesoamerica. ­Goodman24 states chewing, drinking, snufng, and enemas were all common forms of tobacco ingestion in the pre-Contact Americas. A contemporary of Sahagún’s Historia General de las Cosas de Nueva España, the Descripción de San Bartolomé (1588), for example, mentions chewing of pulverized tobacco mixed with lime by the Tz’utujil Maya of highland Guatemala. Te preparation was recog- nized for its strengthening and thirst-quenching properties and kept in small gourds­ 25. More recent ethnographic observations demonstrate the persistence of these consumption and storage patterns for tobacco-based products. ­Winter26 reports Mazatec of Oaxaca consistently carrying powdered green tobacco leaves in little gourds. Te powder is used to relieve fatigue and in witchcraf. Similar practices are observed for the ­Huichol27 and the Tzotzil Maya of Chiapas who call the powder-lime mixture mai28, a term used for tobacco by several Maya sub-groups. In an efort to trace tobacco consumption among the pre-Columbian Maya, Zagorevski and Loughmiller- Newman29 successfully identifed nicotine residues in a ceramic miniature vessel using liquid chromatogra- phy–mass spectrometry (LC–MS) methods. Although there is no provenience information for the fask, the ceramic type situates it in the Southern Maya Lowlands and the Late Classic (AD 550-850) period. Te container is of particular interest, because of a series of painted hieroglyphs reading yotoot may—the home of tobacco­ 30. Te authors point out that the lack of derivatives like cotinine, as well as the seemingly natural levels of nicotine enantiomer ratios, make the presence of thermo-altered tobacco unlikely. In other words, the vessel appears to have held a “fresh” tobacco product rather than its consumed ­remains29. Based on the available ethnohistoric and ethnographic data from larger Mesoamerica, our working hypothesis states that some of the miniature fasks found in the Maya Area contained mixtures of lime (likely ­Ca[OH]2), tobacco (Nicotiana sp.), and other psychoactive and/or aromatic plants. Te creation of an alkaline oral environ- ment assists in the extraction and absorption of alkaloid substances­ 31. Te largely karstic substrate of the Yucatan Peninsula provides for an abundance of raw material and the ancient Maya are known for their advanced lime- based ­industries32. To draw support for our argument, we extracted residues from a collection of pre-Columbian Maya miniature vessels and analyzed them for their chemical composition in comparison to a suite of native plants through a LC–MS platform. Our main objectives include the following:

• Compare archaeological residues to modern plant extracts to increase the possibilities of identifying specifc compounds through the application of a recently established ancient residue metabolomics methods­ 33. • Compare fresh tobacco samples to extracts obtained from cured leaf material to examine the impact of plant processing techniques. • Compare results for curated to recently excavated, untreated vessels to control for the efects of cleansing procedures on the number of extractable residues. • Compare results for distinct vessels regarding form and aperture diameter to control for possible functional diferences among types of containers.

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 2 Vol:.(1234567890) www.nature.com/scientificreports/

Figure 1. Map showing the archaeological sites of artifact recovery (generated via ArcGIS Desktop 10.7.1 [Environmental Systems Research Institute, 2020, Redlands, CA. https://deskt​ op.arcgi​ s.com/en/​ ] using data from Natural Earth).

Materials Archaeological ceramics. Tis study includes data from 14 miniature ceramics. We separate these fasks into two broad categories: recently excavated (untreated) vessels and previously curated vessels (which had been exposed to restorative treatments). Representing the vast majority (n = 12), the second group of fasks was exca- vated during a series of archaeological salvage projects taking place over the past 15 years in and around the city of Mérida, Yucatán, Mexico as part of the Proyecto Arqueológico Región de Mérida (PARME). Even though sub- stantial evidence exists for both Preclassic and Postclassic occupations, the region reached its demographic peak during the Late ­Classic34. Cist burials from this period tend to display an array of grave goods which occasionally include the miniature ceramics under ­study35. Te remaining two untreated artifacts were excavated in 2016 at the site of Ucanha, which has been studied as part of the Proyecto Arqueológico Sacbé Ucí-Cansahcab (PASUC, known as UCRIP in English). Ucanha was frst occupied in the Middle Preclassic (1000-350 BCE) and became a monumental center with over a thousand inhabitants in the Late Preclassic (350 BCE–AD 250) and the Late Classic­ 36,37. Stone causeways connect it to Cansahcab (5 km to the east) and Kancab and Uci (5 km and 13 km to the west, respectively). Both fasks were found during excavations of Structure 65, a large residential platform with megalithic retaining walls, located about 250 m south of the Ucanha monumental core. Ceramic evidence suggests that the residents of Structure 65 were more prosperous than most households at Ucanha. While it cannot be completely ruled out that vessels were touched bare-handed by a present-day tobacco consumer during the excavation process, the large majority of PARME and PASUC project members are non- smokers. Once recovered, artifacts were immediately sealed into labelled plastic bags. It can be expected that the curatorial treatments PARME vessels underwent before their exhibition resulted in the cleansing of exteriors from contaminants deposited during recovery. While exact treatment protocols are unknown, distilled water, water-alcohol solutions, acetone, and xylene are agents commonly used by restoration laboratories of the Instituto Nacional de Antropología e Historia (INAH). Further contamination is unlikely as conservators and laboratory staf wore sterile gloves when handling the artifacts. During exhibition at the Yucatan Regional Museum of Anthropology “Palacio Cantón” PARME vessels were placed in closed vitrines. Moreover, their comparatively small apertures drastically reduce the potential of airborne contamination. Since the exhibition was disassembled, the artifacts have only been removed from sealed containment for the purpose of residue extraction. Based on typological analyses, most of the selected miniature vessels correspond to the Muna ceramic group and the Late to Terminal Classic transition (AD 750–900). One container belongs to the Late Classic Dzitya (AD 600–800) and another one to the Early Classic Aguila (AD 250–600) group. Following Loughmiller-Cardinal and ­Zagorevski20, the collection can be broken down into the following form classes: efgies with narrow apertures (n = 1), paneled fasks with narrow apertures (n = 9), sculpted fasks with narrow apertures (n = 2), and sculpted fasks with wide apertures (n = 2) (Fig. 1). By comparison, the unprovenienced fasks from the Kislak collection analyzed by Loughmiller-Cardinal and Zagorevski all display aperture diameters (2.9–3.2 cm) we would classify

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 3 Vol.:(0123456789) www.nature.com/scientificreports/

V# Proj Prov Group Period Form Decoration Height (cm) Width (cm) Aper (cm) Strung 1 PASUC Ucanha Muna L–T SF None 5 5 0.7 Yes 2 PASUC Ucanha Muna L–T EF Monkey 5.5 5.5 0.8 No 3 PARME NE MID Muna L–T PF Serrated edges 6 7.1 0.8 Yes 4 PARME NE MID Muna L–T PF Painted, Applied buttons 6.9 7.4 0.9 Yes 5 PARME N MID Muna L–T PF None 6.6 7 0.9 Yes 6 PARME N MID Muna L–T PF Painted 6.4 7 0.8 Yes 7 PARME NE MID Muna L–T PF Serrated edges 5.8 6.2 1 Yes 8 PARME NE MID Muna L–T PF Zoomorphic incision 8.2 9.3 1 Yes 9 PARME NE MID Muna L–T PF Applied buttons, negative technique painting, incised 5.6 6.4 0.7 No 10 PARME W MID Muna L–T PF Painted, applied buttons 7.1 7.8 0.9 Yes 11 PARME NE MID Muna L–T PF Serrated edges 7.3 8.3 1.2 Yes 12 PARME NE MID Dzitya L SF Painted, incised 4.8 4.7 1.1 Yes 13 PARME NE MID Aguila E SF Incised 6.4 6.6 2.4 Yes 14 PARME W MID Muna L–T SF Modeled owl face 5.3 5.4 2 Yes

Table 1. PASUC and PARME vessels. NE MID Northeast of Mérida, N MID North of Mérida, W MID West of Mérida, L–T Late-Terminal Classic (AD 750–900), L Late Classic (AD 600–800), E Early Classic (AD 250– 600), PF paneled fask, SF sculpted fask, EF efgy fask.

Species Code Character Distribution Uses Modes Medicinal, religious, stimulant Smoking, Chewing, Snufng, Drink- Nicotiana tabacum (Tobacco) NT Domesticated Across Americas (Winter 2000) ing, Enemas (Goodman 1993) Medicinal, religious, stimulant Smoking, Chewing, Snufng, Drink- Nicotiana rustica (Aztec tobacco) NR Domesticated Across Americas (Winter 2000) ing, Enemas (Goodman 1993) Aromatic, religious, recreational, Smoking, Baths, Drinking (Siegel Tagetes lucida (Mexican marigold) TLU Wild Mexico, Central America medicinal (Siegel et al. 1977) et al. 1977) Nymphaea sp. (Water lily) NSP Wild Across the world Religious (de Rios 1974) Drinking (de Rios 1974) Caribbean, Northern South Lonchocarpus violaceus (Lilac tree) LVI Wild Aromatic (de Lima 1990) Drinking (de Lima 1990) America Smoking, Chewing, Drinking (Val- Salvia divinorum (Diviner’s sage) SDI Wild Western Mexico Religious (Valdés et al. 1983) dés et al. 1983) Medicinal (Stevenson 1915), Reli- Drinking, Chewing (Stevenson Datura wrightii (Sacred datura) DWR Wild, Ornamental Northern Mexico, Southwestern US gious (Baker 2005) 1915) Snufng (Echeverría and Niemeyer Anadenanthera peregrina (Yopo) APE Wild South America Religious (Berenguer 1998) 2013)

Table 2. List of reference species including information on human management and consumption.

as wide. Similar to the Kislak vessels, PASUC and PARME sculpted fasks are consistently smaller in height and width compared to paneled ones. Te latter frequently display applied buttons or serrated edges along their sides. While 60% of the Kislak vessels were designed to be strung as indicated by clay loops and wear marks around the neck, corresponding evidence is present in an even higher percentage (83%) in our collection (Table 1).

Reference species. As mentioned above, the reliance on relatively few biomarkers (i.e., anabasine, caf- feine, nicotine, etc.) limits the ability to confdently determine which plant species are associated with which (type of) artifact. For instance, nicotine is present throughout the Nicotiana genus­ 12. Hence, it does not allow for the discrimination of domesticated (N. tabacum [NT], N. rustica [NR]) from wild tobaccos (e.g., N. attenu- ata, N. quadrivalvis) by ­itself11,33,38. On the other hand, the continuing lack of compound libraries for LC–MS data requires a comparison of archaeological residues with substances produced by reference species in order to attain or approach species identifcations. For this study, we selected a series of modern plant references based on the presence of psychoactive properties and/or their reported association with smoking, snufng, or drinking practices in pre-Columbian Mesoamerica and the Greater Antilles (Table 2). Our hypotheses regarding the presence of these species in our archaeological samples vary. Even though terms for “tobacco” have been deciphered in Maya hieroglyphic writing and cigar smoking is occasionally represented in Maya art­ 39, preserved plant remains are very scarce. Dedrick­ 40 identifed a single Nicotiana seed at the site of K’axob, Belize. As no wild tobacco species are native to the Yucatan ­Peninsula41 and morphological features resemble N. tabacum more than N. rustica (Dedrick, personal communication), it likely stems from the former. Morell-Hart and Cane­ 42–44 recovered a total of seven specimens from four diferent sites in Honduras. Te pho- tography of the Los Naranjos fnd exhibits surface patterning again consistent with N. tabacum42 (Fig. 2). Withal, given their wide-spread distribution, we consider it to be very likely the pre-Columbian Maya had access to both

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 4 Vol:.(1234567890) www.nature.com/scientificreports/

Figure 2. Miniature vessels: Paneled fask with narrow aperture (V11), Sculpted fask with narrow aperture (V12), Sculpted fask with wide aperture (V13).

Figure 3. Clusters established by R Vegan applying Ward’s minimum variance method and the Jaccard index.

species of domesticated tobacco. Assuming the vessels held mixtures of tobacco and aromatic ingredients, we consider Mexican marigold (TLU) to be a strong candidate for a complementary ingredient. On the other hand, we disagree with the idea that waterlily (NSP) was used by the Maya to achieve ASCs. Hence, we do not expect to fnd chemical traces specifc to that species. As L. violaceus bark is employed in alco- holic beverages rather than smoking, snufng, or chewing mixtures, we also predict an absence of diagnostic compounds for the lilac tree (LVI). Te situation is less clear for diviner’s sage (SDI), datura (DWR), and yopo (APE). Ethnohistorical and/or ethnobotanical sources attest to their consumption by neighboring ethnic groups. For yopo, in particular, its presence in the Greater Antilles has been evidenced archaeologically­ 45. Given the long-distance trade networks the Maya were part of during the Classic ­period46, it is not unreasonable to infer the availability of such well-known intoxicating plants. Results Extracted features. Our analysis yielded a list of 9092 mass spectral features in total from the combined collection of vessels and references analyzed. Tese chemical features represent compounds and adducts (and in some cases fragment ions) and isotopomers thereof. Initial cluster analyses (Fig. 3) yielded both expected as well as surprising results. Despite their diferent stages of curing (frozen, cured for 10 days, and cured for 30 days), the three sample types belonging to the two tobacco species group together (clusters in pink and green). Te remaining modern references form a cluster of their own (in black at right). However, the PASUC vessels (in yellow) appear in between the latter and the N. rustica group on a branch which separates them completely from the PARME containers. No further relations based on spatial provenience are recognizable. On the other hand, vessel form seems to have at least some efect on recovered residues. Te cluster formed by Vessels 03 to 09 (in blue) corresponds exclusively to paneled fasks with narrow apertures. Te cluster of vessels including V10 through V14 (in black at lef), however, contains three diferent forms of containers including sculpted fasks with wide apertures. Further examination through successive principal components analysis (PCA) allowed for a more detailed perspective on the variance within the collection as well as similarity/distance between particular (groups of) samples (Fig. 4). Plot A highlights the distance both tobaccos as well as yopo and sacred datura keep with respect to the archaeological samples. Tis means when all 9092 features are considered, the archaeological samples

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 5 Vol.:(0123456789) www.nature.com/scientificreports/

Figure 4. Principal components analysis (PCA) by MetaboAnalyst—(A) Vessels and all references; (B) Vessels, TLU, and LVI; (C) Vessels 03-14 only.

appear to be more similar to the diviner’s sage and the locally collected plant references. Plot B, which only retains the two modern references most similar to the archaeological samples—Mexican marigold (TLU) and the lilac tree (LVI)— shows all archaeological samples scoring quite close to each other on Component 1. However, both PASUC vessels are more distant from their “peers” than TLU. Tis speaks to the degree of variation in residues extracted from curated versus untreated archaeological ceramics. Only afer removing both remaining refer- ences as well as Vessels 01 and 02 is it possible to discern any meaningful patterns among the PARME vessels. Plot C essentially recreates the same clusters observed in Fig. 3 with the only notable diference being Vessel 05’s distance from the rest of the group. Te tight clustering of this larger group of artifacts seems to be directly related to the number of compounds reported per sample. While all modern references with the exception of LVI yielded more than 500 mass spectral features, Fig. 5 exhibits the noticeable quantitative diference between the archaeological vessels. While curated PARME fasks averaged 43 features, the untreated PASUC vessels more than doubled this mark yielding an aver- age of 97 features. In other words, the lower number of recovered residues among the PARME vessels decreases

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 6 Vol:.(1234567890) www.nature.com/scientificreports/

Figure 5. Box plot comparing the number of detected features between curated and untreated vessels.

Species NT10 NR10 TLU DWR APE Metab ID 6956 4225 8738 8776 8791 8813 1739 809 m/z 358.3 349.2 256.3 270.3 312.3 151.0 337.2 339.2 Ret time 11.97 10.22 11.43 11.49 11.38 0.82 11.29 11.71 V01 X X V02 X X X V03 X X X V04 X V05 X V06 X X V07 V08 X X X V09 X X X V10 X X V11 X X V12 X X V13 X X X X X X V14 X X

Table 3. List of unique compounds and their occurrence in archaeological samples.

the discriminatory potential of exploratory statistics such as PCA signifcantly. Tis panorama of homogeneity is likely a case of false negatives.

Shared unique compounds. Te examination of unique compounds found in our reference samples and shared by one or more of the archaeological samples has largely been unsatisfactory. First, we are unable to report a single occurrence of one of the four chemical standards run with the archaeological samples. In other words, we were unsuccessful in repeating the identifcation of nicotine made by Zagorevski and Loughmiller-New- man29 on a similar artifact, despite using instruments of at least comparable (LC–MS) or better (GC-TOFMS) sensitivity. Second, the frequency with which we were able to spot unique compounds in the archaeological samples was generally low. Features unique to SDI, NSP, and LVI were not detected in any of the archaeological samples. Overall, Vessel 13 displays the highest percentage of shared unique compounds: its extracts contained 4 out of 128 compounds exclusively identifed in Mexican marigold (Table 3). Nevertheless, when observing Table 3 several interesting patterns emerge. Te ubiquity of mass spectral feature #6956 is remarkable. Tis single unique compound for N. tabacum, detected exclusively in leaves cured for 10 days, is shared by half of the archaeological samples. N. rustica is also only represented by a single unique compound extracted from the 10-day cured sample (#4225). However, this compound is shared by only one vessel. Together, the NT10 and NR10 samples indicate that although the tobacco references are distant from the archaeological samples when all features are considered, specifc signals might be shared consistently. Te highest overall ubiquity value corresponds to a Mexican marigold compound (#8738). Tis unique compound is shared by 10 of the 14 archaeological samples. Aside from the mass spectral features #6956, #8738, and #8813 (also associated with TLU), the remaining unique compounds fail to be shared by more than 4 out of 14 vessels.

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 7 Vol.:(0123456789) www.nature.com/scientificreports/

Finally, we did not identify any of the shared unique compounds detected in this study; a situation not uncom- mon in non-targeted metabolomics-based analyses. Even though relatively low mass error ranges for m/z peaks allowed us to reconstruct potential molecular formulas for several of those compounds, none of the potential molecules appeared to be present in the KNApSAcK database­ 47. Discussion Although being only the second study of this type, the patterns emerging from the application of ancient residue metabolomics established by Brownstein et al.33 are relevant for future research regarding specifc molecular residue signatures in ancient ceramics. By focusing on the entire set of extractable compounds, rather than a pre- selected suite of biomarkers, we were able to observe that post-recovery treatment does impact chemical residue datasets signifcantly. Restorative practices, directed towards the cleansing and recovery of surface features such as paintings or reliefs, clearly reduce the compounds available for extraction. Tis fact is likely due to the use of diverse solvents during these procedures. By comparison, the lack of association based on spatial provenience in our dataset is encouraging. In other words, the soil matrix surrounding the vessels afer deposition seems to have less of an impact on the results than their contents. Te PASUC vessels strongly support this argument as their metabolomic variance is signifcantly greater when compared to their PARME counterparts, even as the former were recovered from a single structure, while the latter belong to several sites in a larger region. Miniature vessels such as the ones we analyzed most likely held diferent products depending on their general form and, particularly, the size of their apertures. We observed similarity specifcally among paneled fasks with narrow apertures. Te analytical work on shared unique compounds, an approach most similar to the GC–MS focus on specifc biomarkers, led to additional important conclusions. Te frst concerns the characterization and analysis of modern references, which needs to incorporate a larger array of processed products of those candidate species for ancient residues. Humans do not ofen consume plants or animals in a raw state. It is much more likely for organic products to be somehow manipulated before serving their fnal purpose. Te exclusive presence of unique compounds from cured rather than fresh tobacco leaves in our archaeological samples points in this direction. Te second remark pertains to the type of species selected for references. Although our working hypothesis led us to focus on tobaccos and their derivative compounds, the analyses showed us that more evidence might emerge regarding secondary ingredients such as Mexican marigold (T. lucida), an aromatic additive in tobacco mixtures reported by Colonial writers like Friar Bernardino de Sahagún48. In our study, chemical features unique to T. lucida were more commonly detected in archaeological samples than any other targeted substance including the biomarker nicotine. When all signals were considered for analysis, the fasks grouped closer to TLU compared to either NT or NR. Tis is the frst time that Mexican marigold has been detected in archaeological residues, and its association with miniature ceramics contributes to an increasingly nuanced paleoethnobotanical record for Mesoamerica. On the other hand, evidence for the storage of material originating from water lily, lilac tree, diviner’s sage, or yopo is either inconsistent or absent. Te presence of residues associated with Mexican marigold is illuminating for the archaeological study of human use of drugs and cultural practices related to ASCs in general. It provides insight into the persistence in use of certain substances that have maintained importance among native communities in the region. Like other indigenous American cultures, Mesoamerican people like the Nahua or Maya appear to have mixed psychoactive substances with other products­ 21. Almost a century ago, Miner­ 31 called attention to the cross-cultural parallels in the use of lime to enhance the efects of alkaloids contained in tobacco, coca (Erythroxylum coca), or betel (Piper betle). Adams et al.49 suggest tobacco might have been placed inside of Yucca quids in the Southwestern US to reduce gingival irritation or to more evenly spread nicotine release over a longer period of chewing. Siegel et al.21 report that Mexican marigold was used because it “reduces the harshness of yé [N. rustica] ”. In conjunction with ethnohistoric reports of tobacco mixtures, these observations indicate the parapherna- lia surrounding drug consumption extends beyond tools for preparation, delivery devices, and containers. A holistic understanding of entheogenic drug complexes ought to incorporate the non-psychoactive ingredients of the ingested products. Further interdisciplinary research is necessary to frst detect these elements and then identify their roles in achieving ASCs. Biomolecular archaeology and ancient residue metabolomics clearly can add signifcantly to such research and discussions about the role of psychoactive plants throughout the world. Methods Modern plant references. NT and DWR plants were grown in WSU greenhouses from seed acquired through Horizon Herbs Strictly Medicinal (Williams, OR, USA). Similarly, NR seeds were acquired through the United States Department of Agriculture (USDA) Agricultural Research Services (ARS) National Plant Germplasm System (NPGS) (PI 555554). SDI cuttings were originally taken in the Sierra Mazateca mountains of Oaxaca, Mexico, and then also propagated in WSU greenhouses. APE seeds were ordered through Natural Ether (Coral Springs, FL, USA) and ground in order to simulate the reported snufng mode of administration (e.g., Ogalde et al. 2009). TLU leaves and LVI bark shavings were collected on the Campus de Ciencias Sociales, Económico-Administrativas y Humanidades of the Universidad Autónoma de Yucatán (UADY), Mexico and NSP fowers and stalk material on the Campus de Ciencias Biológicas y Agropecuarias, also of UADY, Mexico. Te preparation of reference samples proceeded along the following protocol: With the exception of both Nicotiana representatives, one sample was prepared for each candidate species using dried leaves (TLU, SDI), stem and fower (NSP), rind (LVI), or ground seeds (DWR, APE). For both tobaccos, a sample of freeze-dried leaves (NTFD; NRFD) was compared to material cured for 10 (NT10; NR10) and 30 days (NT30; NR30), respec- tively. In addition to references created directly from modern plants, chemical standards for LC–MS analysis of

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 8 Vol:.(1234567890) www.nature.com/scientificreports/

nicotine, as well as atropine, tropine, and scopolamine, which are found in Datura species, were purchased from Sigma-Aldrich (St. Louis, MO, USA).

Extraction and chromatography. Vessels were emptied of deposited sediment in advance of residue extraction for metabolomic analysis. Disposable wooden sticks were used to pry dry soil from fask interiors. Te resulting samples were used for semi-quantitative chemical tests and macrobotanical analysis, which did not yield evidence for tobacco or other plant species­ 50. Te extraction protocol included the successive exposure of the interior surfaces of all 14 vessels to three diferent solvent systems: (1) 2% aqueous tartaric acid (TA), (2) acetonitrile:2-propanol:water [3:2:2] (APW), and methyl tert-butyl ether (MTBE). We followed this approach to increase our ability of capturing compounds with diferent polarity characteristics. Each vessel was flled with 10 ml of TA and then sonicated for 10 min. Aferwards, the supernatant was transferred to a clean vial and the extraction was repeated with the other two solvents, APW second and MTBE last. Sonication has proven to consistently extract chemical residues from the matrix of artifacts made from clay and porous ­rocks51, thereby preventing the reliance on samples originating from surface layers which have been shown to be susceptible to airborne ­contamination52. Sonication also eliminates the need for supplies used to clean or abrade artifact sur- faces such as cotton balls or sandpaper, which have similarly been acknowledged as vectors of contamination­ 53. Nonetheless, to control for possible intrusive substances, blank samples were created for all three solvents. Blanks were collected in the same type of vial used for the archaeological samples and from thereon processed identically. Tis procedure guarantees that all possible contamination stemming from supplies or instruments can be accounted for during data processing. Depending on the solvent system, diferent techniques were then applied to convert extracts to a solid state: TA samples were freeze-dried; APW samples rota-evaporated and freeze-dried; and MTBE samples air-dried in a fume hood. Te protocol was identical for the modern plant references. In addition, blank samples were run for each solvent. All solvents used for extraction and analyses were of liquid chromatography-mass spectrometry grade. Resuspension occurred following the protocol previously employed by Brownstein et al.33. We used 0.6 ml of 0.1% formic acid/water:acetonitrile [1:1] for TA and APW extracts and an identical volume of 0.10% formic acid/acetonitrile for MTBE extracts. Resuspended extracts were vortexed and centrifuged at 10,000×g for 10 min at 4 °C, before a 0.5 ml aliquot was transferred to liquid chromatography sample vials. Similarly, instrument confguration resembled the setting published by Brownstein et al.33. Ultra-performance liquid chromatography (UPLC) was conducted on a Waters ACQUITY UPLC system (Waters Corporation, Milford, MA, USA) with photodiode array (PDA) detection ranging between 210 and 400 nm followed by inline mass spectrometric analysis. One microliter of sample was injected through a 2.0 µL sample loop using the full loop injection mode, and the fow rate through a Waters ACQUITY UPLC T3 column (HSS T3, 1.8 µm, 2.1 × 100 mm) was 0.32 mL/min with 0.10% formic acid/water (A) and 0.10% formic acid/methanol:acetonitrile [2:3] (B) in a slightly concave gradient elution mode. Te gradient elution was applied as follows—97% A:3% B to 15% A:85% B from 0.00 to 12.00 min, changed to 3% A:97% B in 0.10 (12.10) min, maintained at 3% A:97% B until 14.00 min, returned to the initial conditions of 97% A:3% B in 0.10 (14.10) min, and then, before the next injection, maintained at 97% A:3% B until 16.00 min. Te analysis time was 16.00 min. Te autosampler chamber and column temperature were 8 °C and 35 °C, respectively. A Waters SYNAPT G2-S HDMS Q-TOF with lockspray ionization was operated in ESI positive and resolution mode (mass resolution of ~ 40,000 for the compound range analyzed). Te scan range was from 100 to 1200 m/z with a scan time of 0.3 s. Mass spectral data were collected in profle mode using MS­ E with high collision energy (ramp 15–40 V) for fragmentation. Te capillary voltage, sampling cone voltage, and source ofset voltage were 3.0 kV, 60 V, and 60 V, respectively. Te source temperature was 100 °C with a cone gas (nitrogen) fow rate of 50 L h−1. Te desolvation temperature was 250 °C with a desolvation gas (nitrogen) fow rate of 900 L h−1. Te nebulizer gas (nitrogen) fow was 6.0 bar and the lock mass compound was leucine enkephalin with a reference mass of 556.2771 [M + H]+ m/z33.

Data processing. Although Waters LC–MS systems are regularly used in association with Progenesis QI sofware, our team extended data processing and multivariate statistical analyses to the application of a series of diferent open-source tools including MZmine ­254, ­MetaboAnalyst55, Draw Venn ­Diagram56, and the “vegan” package for ­R57. Te detailed data processing method is described in Brownstein et al.33. However, before carry- ing out any statistical procedure, the list of mass spectral features was reduced by all compounds encountered in solvent blanks. All datasets were submitted to the ­MetaboLights58 repository (study identifer: MTBLS2018).

Received: 2 September 2020; Accepted: 17 December 2020

References 1. Fitzpatrick, S. M. Ancient psychoactive substances (University Press of Florida, Gainesville, 2018). https​://doi.org/10.2307/j.ctvx0​ 76mt. 2. Winkelman, M. J. Shamans and other “magico-religious” healers: a cross-cultural study of their origins, nature, and social trans- formations. Ethos 18(3), 308–352 (1990). 3. Carod-Artal, F. J. Hallucinogenic drugs in pre-columbian mesoamerican cultures. Neurol. (English Ed.) 30(1), 42–49. https​://doi. org/10.1016/j.nrlen​g.2011.07.010 (2015). 4. Torres, C. Archaeological evidence for the antiquity of psychoactive plant use in the Central Andes. Ann. Mus. Civ. di Rovereto 1995(11), 291–326 (1995). 5. Frecska, E. & Kulcsar, Z. Social bonding in the modulation of the physiology of ritual trance. Ethos 17(1), 70–87 (1989).

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 9 Vol.:(0123456789) www.nature.com/scientificreports/

6. Rush, J. A. Entheogens and the development of culture: the anthropology and neurobiology of ecstatic experience: essays (North Atlantic Books, Berkeley, 2013). 7. Hagen, E. H. & Tushingham, S. Te prehistory of psychoactive drug use. In Handbook of cognitive archaeology (eds Henley, T. B. et al.) 471–498 (Routledge, New York, 2019). https​://doi.org/10.4324/97804​29488​818-25. 8. McClenon, J. Shamanic healing, human evolution, and the origin of religion. J. Sci. Study Relig. 36(3), 345–354 (1997). 9. La Barre, W. Old and new world narcotics: a statistical question and an ethnological reply. Econ. Bot. 24(1), 73–80 (1970). 10. Raferty, S. M. Prehistoric intoxicants of North America. In Ancient psychoactive substances (ed. Fitzpatrick, S. M.) 112–127 (Uni- versity Press of Florida, Gainesville, 2018). 11. Tushingham, S., Snyder, C. M., Brownstein, K. J., Damitio, W. J. & Gang, D. R. Biomolecular archaeology reveals ancient origins of indigenous tobacco smoking in North American plateau. Proc. Natl. Acad. Sci. U. S. A. 115(46), 11742–11747. https​://doi. org/10.1073/pnas.18137​96115​ (2018). 12. Sisson, V. & Severson, R. Alkaloid composition of the Nicotiana species. Beiträge zur Tab. Int. to Tob. Res. 14(6), 327–339. https://​ doi.org/10.2478/cttr-2013-0610 (1990). 13. Clement, B. A., Gof, C. M. & Forbes, T. D. A. Toxic amines and alkaloids from acacia rigidula. Phytochemistry 49(5), 1377–1380. https​://doi.org/10.1016/S0031​-9422(97)01022​-4 (1998). 14. Holdsworth, D. K., Jones, R. A. & Self, R. Volatile alkaloids from areca catechu. Phytochemistry 48(3), 581–582 (1998). 15. Janiger, O. & de Rios, M. D. Suggestive hallucinogenic properties of tobacco. Med. Amthropol. Newsl. 4(4), 6–11 (1973). 16. Groark, K. P. Te angel in the gourd: ritual, therapeutic, and protective uses of tobacco (Nicotiana Tabacum) among the Tzeltal and Tzotzil Maya of Chiapas, Mexico. J. Ethnobiol. 30(1), 5–30. https​://doi.org/10.2993/0278-0771-30.1.5 (2010). 17. Stewart, G. A history of the medicinal use of tobacco 1492–1860. Med. Hist. 11(3), 228–268 (1967). 18. Schultes, R. E. Plants of the gods: origins of hallucinogenic use (McGraw-Hill, New York, 1979). 19. de la Garza, M. Sueños y Éxtasis: Visión Chamánica de Los Nahuas y Mayas (Fondo de Cultura Economica, Mexico City, 2012). 20. Loughmiller-Cardinal, J. A. & Zagorevski, D. Maya fasks: the home of tobacco and godly substances. Anc. Mesoam. 27(1), 1–11. https​://doi.org/10.1017/S0956​53611​60000​79 (2016). 21. Siegel, R. K., Collings, P. R. & Diaz, J. L. On the use of Tagetes lucida and Nicotiana rustica as a Huichol smoking mixture: the Aztec “Yahutli” with suggestive hallucinogenic efects. Econ. Bot. 31(1), 16–23. https​://doi.org/10.1007/BF028​60647​ (1977). 22. Kaplan, L. Historical and ethnobotanical aspects of domestication in tagetes. Econ. Bot. 14(3), 200–202. https​://doi.org/10.1007/ BF029​07950​ (1960). 23. Robicsek, F. Ritual smoking in central America. In Smoke: a global history of smoking (eds Gilman, S. L. & Zhou, X.) 30–37 (Reak- tion Books, London, 2004). 24. Goodman, J. Tobacco in history (Routledge, London, 1993). 25. Tompson, J. E. Maya history and religion (University of Oklahoma Press, Norman, 1970). 26. Winter, J. C. Food of the gods: biochemistry, addiction, and the development of native American tobacco use. In Tobacco use by native North Americans: sacred smoke and silent killer (ed. Winter, J. C.) 305–330 (University of Oklahoma Press, Norman, 2000). 27. Myerhof, B. G. Peyote hunt. Te sacred journey of the Huichol Indians (Cornell University Press, Ithaca, 1974). 28. Starr, F. Notes upon the ethnography of Southern Mexico (Putnam Memorial Publication Fund, Davenport, 1900). 29. Zagorevski, D. V. & Loughmiller-Newman, J. A. Te detection of nicotine in a Late Mayan period fask by gas chromatography and liquid chromatography mass spectrometry methods. Rapid Commun. Mass Spectrom. 26, 403–411. https​://doi.org/10.1002/ rcm.5339 (2012). 30. Stuart, D. Te inscriptions from temple XIX at Palenque: a commentary (Te Pre-Columbian Art Research Institute, Washington, 2005). 31. Miner, H. Parallelism in alkaloid-alkali quids. Am. Anthropol. 41(4), 617–619 (1939). 32. Seligson, K. E., Ortiz Ruiz, S. & Barba Pingarrón, L. Prehispanic maya burnt lime industries: previous studies and future directions. Anc. Mesoam. 30(2), 199–219. https​://doi.org/10.1017/S0956​53611​70003​47 (2019). 33. Brownstein, K. J., Tushingham, S., Damitio, W. J., Nguyen, T. & Gang, D. R. An ancient residue metabolomics-based method to distinguish use of closely related plant species in ancient pipes. Front. Metab. 7, 133 (2020). 34. Pantoja Díaz, L., Medina Martín, C. & Gómez Cobá, M. J. San Pedro Cholul: Un Asentamiento Arqueológico Del Clásico Tardío En La Región de Mérida, Yucatán, México. In Te archaeology of Yucatan (ed. Stanton, T. W.) 165–180 (Archaeopress, Oxford, 2014). 35. Medina Martín, C., Gómez Cobá, M. J. & Pantoja Díaz, L. R. Las Costumbres Funerarias En La Periferia de Mérida, Yucatán. In Te archaeology of Yucatan (ed. Stanton, T. W.) 181–196 (BAR Publishing, Oxford, 2014). 36. Hutson, S. R. & Welch, J. A. Sacred landscapes and building practices at Uci, Kancab, and Ucanha, Yucatan, Mexico. Anc. Mesoam. 25(2), 421–439. https​://doi.org/10.1017/S0956​53611​40002​85 (2014). 37. Kidder, B. Negotiating household quality of life and social cohesion at Ucanha, Yucatan, Mexico, during the late preclassic to early classic transition. University of Kentucky (2019). 38. Tushingham, S. & Eerkens, J. W. Hunter-gatherer tobacco smoking in ancient North America: current chemical evidence and a framework for future studies. In Perspectives on the archaeology of pipes, tobacco and other smoke plants in the ancient Americas (eds Bollwerk, E. & Tushingham, S.) 211–230 (Springer, New York, 2016). 39. Cagnato, C. Shedding light on the nightshades (solanaceae) used by the ancient Maya: a review of existing data, and new archeo- botanical (macro- and microbotanical) evidence from archeological sites in Guatemala. Econ. Bot. 72(2), 180–195. https​://doi. org/10.1007/s1223​1-018-9412-8 (2018). 40. Dedrick, M. Te distributed household: plant and mollusk remains from K’axob (University of North Carolina, Belize, 2014). 41. Arellano Rodríguez, J. A., Flores Guido, J. S., Tun Garrido, J., & Cruz Bojórquez, M. M. Nomenclatura, Forma de Vida, Uso, Manejo y Distribución de Las Especies Vegetales de La Península de Yucatán. UADY: Mérida (2003). 42. Morell-Hart, S., Joyce, R. A. & Henderson, J. S. Multi-proxy analysis of plant use at formative period los Naranjos, Honduras. Lat. Am. Antiq. 25(1), 65–81 (2014). 43. Morell-Hart, S. Paradigms and syntagms of ethnobotanical practice in pre-hispanic Northwestern Honduras (University of California, Berkeley, 2011). 44. Morell-Hart, S., Joyce, R. A., Henderson, J. S. & Cane, R. Ethnoecology in pre-hispanic central America: foodways and human- plant interfaces. Anc. Mesoam. 30(3), 535–553. https​://doi.org/10.1017/S0956​53611​90000​14 (2019). 45. Pagán-Jiménez, J. R. & Carlson, L. A. Recent archaeobotanical fndings of the hallucinogenic snuf cojoba (Anadenanthera Per- egrina (L.) SPEG.) in precolonial Puerto Rico. Lat. Am. Antiq. 25(1), 101–116 (2014). 46. Sharer, R. & Traxler, L. P. Te ancient Maya (Stanford University Press, Stanford, 2006). 47. Afendi, F. M. et al. KNApSAcK family databases: integrated metabolite-plant species databases for multifaceted plant research. Plant Cell Physiol. 53(2), e1–e1. https​://doi.org/10.1093/pcp/pcr16​5 (2012). 48. de Sahagún, F. B. Historia General de Las Cosas de Nueva España [1569] (Fondo de Cultura Economica, Mexico City, 1999). 49. Adams, K. R., Johnson, K. L. & Murphy, T. M. Prehistoric Puebloan yucca (Yucca) quids with wild tobacco (Nicotiana) contents: molecular and morphological evidence from Antelope Cave, northwestern Arizona. J. F. Archaeol. 40(3), 310–324. https​://doi. org/10.1179/20424​58214​Y.00000​00006​ (2015). 50. Zimmermann, M. Informe Final Del Análisis de Tres Vasijas Cerámicas Miniatura; Mérida (2013).

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 10 Vol:.(1234567890) www.nature.com/scientificreports/

51. Eerkens, J. et al. GC/MS analysis of residues reveals nicotine in two late prehistoric pipes from CA-ALA-554. Proc. Soc. Calif. Archaeol. 26, 212–219 (2012). 52. Washburn, D. K., Washburn, W. N., Shipkova, P. A. & Pelleymounter, M. A. Chemical analysis of cacao residues in archaeologi- cal ceramics from North America: considerations of contamination, sample size and systematic controls. J. Archaeol. Sci. 50(1), 191–207. https​://doi.org/10.1016/j.jas.2014.07.011 (2014). 53. King, A., Powis, T. G., Cheong, K. F. & Gaikwad, N. W. Cautionary tales on the identifcation of cafeinated beverages in North America. J. Archaeol. Sci. 85, 30–40. https​://doi.org/10.1016/j.jas.2017.06.006 (2017). 54. Pluskal, T., Castillo, S., Villar-Briones, A. & Orešič, M. MZmine 2: modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profle data. BMC Bioinform. 11(1), 395. https​://doi.org/10.1186/1471-2105-11-395 (2010). 55. Xia, J., Sinelnikov, I. V., Han, B. & Wishart, D. S. MetaboAnalyst 3.0—making metabolomics more meaningful. Nucleic Acids Res. 43(W1), W251–W257. https​://doi.org/10.1093/nar/gkv38​0 (2015). 56. Van de Peer, Y. Draw venn diagram: calculate and draw custom venn diagrams. http://bioin​forma​tics.psb.ugent​.be/webto​ ols/%0AVen​n/. 57. Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P. R., O’Hara, R. B., Simpson, G. L., Solymos, P., Stevens, M. H. H., Szoecs, E., & Wagner, H. vegan: community ecology package. R package version 2.5-1. https​:// cran.r-proje​ct.org/web/packa​ges/vegan​/index​.html. 58. Haug, K. et al. MetaboLights—an open-access general-purpose repository for metabolomics studies and associated meta-data. Nucleic Acids Res. 41(D1), D781–D786. https​://doi.org/10.1093/nar/gks10​04 (2013). Acknowledgements We thank the Instituto Nacional de Antropología e Historia for permission to carry out the research. Funding for this project was provided by the National Science Foundation (Grant numbers 1419506 and 1918966), as well as PARME and PASUC. Tis work was also supported in part by the USDA National Institute of Food and Agriculture, Hatch project 227700 (DRG). We extend our gratitude towards the staf of restorers at Museo Palacio Cantón for the great care with which all PARME vessels were handled during curation and exhibition. In addition, we are in debt with Dr. Sergio Rodríguez Morales and his staf at the Laboratory for Natural Marine Resources, Dept. of Chemistry, UNAM, at Sisal, for providing access to his rotary evaporator and lyophilizing equipment. We also want to express our gratitude to Dr. Anna Berim for reviewing our metabolomic datasets and providing important insight. Finally, we thank two anonymous reviewers for their insightful comments and constructive critiques of our manuscript. Authors’ contributions Conceptualization MZ, KJB, ST, Data curation KJB, DRG, Formal Analysis KJB, MZ, Funding acquisition ST, DRG, Investigation MZ, IAA, LPD, SH, BK, Methodology, KJB, ST, DRG, Project administration ST, DRG, LPD, Resources ST, DRG, LPD, SH, Sofware, KJB, Supervision, ST, DRG, Validation ST, DRG, Visualization MZ, KJB, IAA, BK, Writing—original draf, MZ, KJB, Writing—review and editing ST, DRG, SH.

Competing interests Te authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to M.Z. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. Open Access Tis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/.

© Te Author(s) 2021

Scientifc Reports | (2021) 11:1590 | https://doi.org/10.1038/s41598-021-81158-y 11 Vol.:(0123456789)